Package ClamR. R topics documented: July 29, Type Package

Transcription

1 Type Package Package ClamR July 29, 2015 Title Time Series Modeling for Climate Change Proxies Version Date Imports stats Author Jonathan M. Lees Maintainer Jonathan M. Implementation of the Wilkinson and Ivany (2002) approach to paleoclimate analysis, applied to isotope data extracted from clams. License GPL LazyLoad yes NeedsCompilation no Repository CRAN Date/Publication :21:30 R topics documented: ClamR-package CLAM climate elliot_yr elliot_yr error.bar NextPow otolith plotproxy.error proxya proxyjk RESCALE rwp_limpet SinMod windowsize

2 2 ClamR-package Index 20 ClamR-package Climate Change with Proxies Details This program implements and improves upon the Wilkinson and Ivany approach to climate time series modeling. The jackknife is used to estimate the 95 percent confidence bounds for the modeled estimates. dx should be chosen to be approximately half a cycle or more. Package: ClamR Type: Package Version: 1.0 Date: License: GPL LazyLoad: yes Author(s) Jonathan M. Lees Maintainer: Jonathan M. Lees<jonathan.lees@unc.edu> Wilkinson, B. H. and Ivany, L. C., Paleoclimatic inference from stable isotope profiles of accretionary biogenic hardparts; a quantitative approach to the evaluation of incomplete data, Palaeogeography, Palaeoclimatology, Palaeoecology, vol. 185, no. 1-2, pp , 01 Sep doi: /j.palaeo ## Not run: data(clam1) x = CLAM1$x y = CLAM1$y dx = gout = proxyjk(x, y, dx)

3 CLAM1 3 plotproxy1(x, y, gout) par(mfrow=c(2,1)) plotproxy.error(x, y, gout, type = 1) plotproxy.error(x, y, gout, type = 2) par(mfrow=c(2,1)) plotproxy.error(x, y, gout, type = 2) plotproxy.all2(gout,yaxstyle=1 ) ## End(Not run) CLAM1 Clam Proxy Data Proxy data from the Orknay Islands data(clam1) Format The format is: List of 2 $ x: num [1:91] $ y: num [1:91] Details Duplicate data has been removed. Source One-year data interval ( mm) of δ 18O record of an archaeological limpet Patella vulgata (specimen QG ) from Orkney, Scotland (Surge and Barrett, 2012). doi: /j.palaeo

4 4 climate data(clam1) ## maybe str(clam1) ; plot(clam1)... plot(clam1$x, CLAM1$y, type="b", xlab="distance", ylab="d18o" ) ## Not run: ###### this is an example from Wang et al.: #### it takes too long to run on CRAN, but should work shellx=clam1$x[38:70] shelly=clam1$y[38:70] window_shell=windowsize(shellx,shelly,1.8,9.4,0.2) #the window size is 5mm, and make all the plots together gout_shell = proxyjk(shellx, shelly, 5) par(mfrow=c(3,2)) plot(shellx,shelly,type="b", xlab="distance from Margin (mm)", ylab=expression(delta*"18o(ppm VPDB)"), xlim=c(4,10), ylim = c(1.5,4)) plot((window_shell$win)/2,window_shell$error,xlab="window Size (mm)", ylab="error", xlim=c(1.6/2,9.4/2), ylim=c(0,0.5)) abline(v=4.63/2, lty="dotdash",col="black") abline(v=5/2, col="black") plotproxy1(shellx, shelly, gout_shell, xlim=c(4,10), ylim = c(1.5,4), xlab="distance from Margin (mm)", ylab=expression(delta*"18o(ppm VPDB)"), main="") plotproxy.all(gout_shell,yaxstyle=1, xlim=c(4,10), ylim1=c(0,4), ylim2=c(-15,5)) plotproxy.error(shellx, shelly, gout_shell, type = 1, xlim=c(4,10), ylim = c(1.5,4), xlab="distance from Margin (mm)", ylab=expression(delta*"18o(ppm VPDB)")) plotproxy.error(shellx, shelly, gout_shell, type = 2, xlim=c(4,10), ylim = c(1.5,4), xlab="distance from Margin (mm)", ylab=expression(delta*"18o(ppm VPDB)")) ## End(Not run) climate Climate Record At Croig Cave Application to modern climate record at Croig Cave and make comparison between reconstructed temperatures and instrumentally measured temperatures.

5 elliot_yr1 5 data("climate") Format A data frame with 360 observations on the following 3 variables. Month a numeric vector overall a numeric vector Temperature a numeric vector Source Monthly sea surface temperature (SST) record for the years derived from observations near Croig Cave, an archaeological site on the Isle of Mull in the Hebrides Islands west of mainland Scotland (Extended Reconstructed Sea Surface Temperature, Smith and Reynolds, 2004). doi: /j.palaeo data(climate) climate_month <- climate$overall climate_temp <- climate$temperature plot(climate_month,climate_temp,type="l", xlab="month",ylab=expression(paste("temperature ("^"o","c)"))) elliot_yr1 Elliot Data Summer Two years of data from Elliot data("elliot_yr1")

6 6 elliot_yr2 Format The format is: List of 4 $ date1 : num [1:28] $ d18o1 : num [1:28] $ date_temp1: num [1:31] $ d18o_pred1: num [1:31] Details Data consists of date, δ18o, temperature and predicted anomaly for two years of data. Source Data sets are selected from the δ18o record of a modern Mercenaria mercenaria shell collected live from Cedar Key in northern Florida and analyzed by Elliot et al. (2003). Series elliot_yr1 records one summer (including the most negative δ18o). Series elliot_yr2 records one winter (including the most positive δ18o). Because the modern Mercenaria mercenaria shell by Elliot et al. (2003) is well dated and its in situ records of SST and salinity are available, the predicted δ18o are also derived from the local instrumental data. Temp is the predicted δ18o for the summer interval of Year1 and Temp2 is the predicted δ18o for the winter interval of Year2. doi: /j.palaeo data(elliot_yr1) plot(elliot_yr1$date1,elliot_yr1$d18o1,xlab="age(years)", ylab=expression(delta* 18O(ppm VPDB) ), xlim=c(1994.9,1996.2), ylim=c(-2.5,2)) elliot_yr2 Elliot Data Winter Winter season of data from Elliot data. data("elliot_yr2")

7 error.bar 7 Format Details Source The format is: List of 4 $ date1 : num [1:28] $ d18o1 : num [1:28] $ date_temp1: num [1:31] $ d18o_pred1: num [1:31] Data consists of date, δo18, temperature and predicted data for winter season of data. See explanation in elliot_yr1. doi: /j.palaeo data(elliot_yr2) plot(elliot_yr2$date2,elliot_yr2$d18o2,xlab="date(year)", ylab=expression(delta* 18O(ppm VPDB) ), xlim=c(1995.2, ), ylim=c(-2.5,1.8)) error.bar Error bar plot Make an X-Y plot with error bars. error.bar(x, y, lo, hi, pch = 1, col =1, barw = 0.1, add = FALSE,...) Arguments x y lo hi pch X-values Y-values Lower limit of error bars Upper limit of error bars plotting character

8 8 NextPow2 col color barw width of the bar add logical, add=false starts a new plot... other plotting parameters Value graphical side effects Author(s) Jonathan M. Lees<jonathan.lees@unc.edu> x = 1:10 y = 2*x+5 zup = rnorm(10) zup = zup-min(zup)+.5 zdown = rnorm(10) zdown = zdown-min(zdown)+.2 #### example with same error on either side: error.bar(x, y, y-zup, y+zup, pch = 1, col = brown, barw = 0.1, add = FALSE) #### example with different error on either side: error.bar(x, y, y-zdown, y+zup, pch = 1, col = brown, barw = 0.1, add = FALSE) NextPow2 Next power of 2 Find the next integer power of 2 NextPow2(x) Arguments x integer Value integer that is a power of 2 higher than given integer

9 otolith 9 Author(s) Jonathan M. Lees<jonathan.lees@unc.edu> NextPow2(600) NextPow2(1023) NextPow2(1025) otolith otolith Proxy Data Early Oligocene otolith from the US Gulf Coast. data(otolith) Format A data frame with 63 observations on the following 2 variables. distance a numeric vector d18o a numeric vector Details Duplicate data have been removed. Source δ18o record of an aragonite otolith from the early Oligocene Rosefield Clay in the US Gulf Coast (Ivany, 2000). doi: /j.palaeo \

10 10 plotproxy.error data(otolith) plot(otolith$distance, otolith$d180) plotproxy.error Plot Output Jack-Knife Plot output of proxyjk, the jackknife estimate of the time series analysis fitting curves. plotproxy.error(x,y,gout, type=1, xlim=null, ylim=null, ylab="", xlab="", main="" ) plotproxy.all(gout, ylab1="", ylab2="",xlab="", main="", xlim=null, ylim1=null, ylim2=null, legposition="topleft", YAXstyle=0, pbox=true, legnames = c( Phs, Pos, Amp, Prd ) ) plotproxy1(x, y, gout, xlim = NULL, ylim = NULL, ylab = "", xlab = "", main = "") plotproxy.error11(x, y, gout, type = 1, xlim = NULL, ylim = NULL, ylab = "", xlab = "", main = "") plotproxy.all2(gout, ylab1 = "", ylab2 = "", xlab = "", main = "", xlim = NULL, ylim1 = NULL, ylim2 = NULL, legposition = "topleft", YAXstyle = 0, pbox = TRUE, legnames = c("phs", "Pos", "Amp", "Prd")) Arguments x y gout type xlim ylim original x values from file original y values from file output of proxyjk type of error bar plotting: 1 = bars, 2=shaded 2-vector(limit on x-axis) 2-vector(limit on xy-axis)

11 plotproxy.error 11 ylim1 ylim2 ylab ylab1 ylab2 xlab main legposition YAXstyle pbox legnames 2-vector(limit on xy-axis) 2-vector(limit on xy-axis) character, y-axis label character, y-axis label character, y-axis label character, x-axis label character, title label legend position Style for Y-axis logical names for legend Details Value Takes output directly from program graphical side effects Author(s) Jonathan M. doi: /j.palaeo See Also proxyjk ## Not run: ## example to read in your data from csv file (for non ClamR data ## fn = "donna_viking_1.csv" ## C1 = scan(file=fn, what=list(mm="", o18=""), sep=",") ## x = as.numeric(c1$mm) ## y = as.numeric(c1$o18) ## x = x[!is.na(y)] ## y = y[!is.na(y)]

12 12 proxya data(clam1) x = CLAM1$x y = CLAM1$y dx = gout = proxyjk(x, y, dx) plotproxy1(x,y,gout) ## End(Not run) proxya Optimum Wilkinson Curve Fitting Runs one cycle of optimum Wilkinson curve fitting for a single sinusoid fitting. proxya(ax, ay, xin) Arguments ax ay xin x-axis values y-axis values starting model: c( Phs,Pos,Amp,Prd) Details Value Note This program implements the Wilkinson and ivany approach to climate time series modeling. This is used in the more sophisticated proxyjk code. Optimum model, vector of 4 values Uses stats package routine optim for optimization Author(s) Jonathan M. Lees<jonathan.lees@unc.edu>

13 proxyjk 13 doi: /j.palaeo \ Wilkinson, B. H. and Ivany, L. C., Paleoclimatic inference from stable isotope profiles of accretionary biogenic hardparts; a quantitative approach to the evaluation of incomplete data, Palaeogeography, Palaeoclimatology, Palaeoecology, vol. 185, no. 1-2, pp , 01 Sep 2002 See Also proxyjk data(clam1) x = CLAM1$x y = CLAM1$y dx = Aout = proxya(x, y, dx) proxyjk Jackknife Wilkinson Curve Fitting Perform a jackknife estimate of proxy curve fitting for time series analysis. proxyjk(x, y, dx) Arguments x y dx x-axis values y-axis values width of window to span in time Details Routine that improves on the Wilkinson and Ivany(2002) approach to climate time series modeling. The jackknife is used to estimate the 95 percent confidence bounds for the modeled estimates. dx should be chosen to be approximately half a cycle or more.

14 14 proxyjk Value List: OUT list( par, mid, ax, predmid,jkest, JKvar, PSTILDE ) omids output midpoints pmids values at output midpoints x input x y input y predy predicted y from spline fit Note See proxya for a duplication of the Wilkinson codes. Author(s) Jonathan M. Lees<jonathan.lees@unc.edu> doi: /j.palaeo See Also proxya ## Not run: ########## this is for reading in data ######## fn = "/home/lees/donna/donna_viking_1.csv" ## fn = "donna_viking_1.csv" ######## C1 = scan(file=fn, what=list(mm="", o18=""), sep=",") ######## x = as.numeric(c1$mm) ######## y = as.numeric(c1$o18) ########x = x[!is.na(y)] ########y = y[!is.na(y)] data(clam1) x = CLAM1$x y = CLAM1$y

15 RESCALE 15 dx = gout = proxyjk(x, y, dx) plotproxy1(x, y, gout) par(mfrow=c(2,1)) plotproxy.error(x, y, gout, type = 1) plotproxy.error(x, y, gout, type = 2) par(mfrow=c(2,1)) plotproxy.error(x, y, gout, type = 2) plotproxy.all2(gout, YAXstyle=1 ) ## End(Not run) RESCALE Rescale a vector to fit in a certain range Rescale a vector to fit in a certain range RESCALE(x, nx1, nx2, minx, maxx) Arguments x nx1 nx2 minx maxx vector new minimum new maximum old min old max Details Used for graphics. Value scale vector is returned

16 16 rwp_limpet Author(s) Jonathan M. Lees<jonathan.lees.edu> x = rnorm(10) RESCALE(x, 3, 9, min(x), max(x) ) rwp_limpet Reconstructed Temperature Record Format Source Application to estimated temperatures from archaeological RWP shell 103a-39-1 with 3 years of temperature data. data("rwp_limpet") A data frame with 74 observations on the following 8 variables. distance_all a numeric vector temp_all a numeric vector distance1 a numeric vector temp1 a numeric vector distance2 a numeric vector temp2 a numeric vector distance3 a numeric vector temp3 a numeric vector RWP shell 103a-39-1 doi: /j.palaeo data(rwp_limpet) plot(rwp_limpet$distance_all, rwp_limpet$temp_all, type= b )

17 SinMod 17 SinMod Sine Model of climate time series Sine Model of climate time series SinMod(x, myex, dc2) Arguments x myex dc2 input model x consisting of 4 values, phase, position, amplitude, period externally defined X-values externally defined observations at X Details This is the function used in optimization of sinusoidal fits to climate data. Value squared sum of difference between obsereved and predicted Note Uses stats package for optimization Author(s) Jonathan M. Lees<jonathan.lees@unc.edu> doi: /j.palaeo See Also optim, proxya, proxyjk

18 18 windowsize data(clam1) x = CLAM1$x y = CLAM1$y dx = A1 = proxya(x, y, dx) RMSout = SinMod(A1$par, x, y) windowsize Find Window Size Estimate optimal window size for seasonal time series analysis. windowsize(x, y, winmin, winmax, winstep) Arguments x y winmin winmax winstep original x values from file original y values from file Minimum window size Maximum window size step size Value win error optimum window length error for win Author(s) Jonathan M. Lees<jonathan.lees@unc.edu> doi: /j.palaeo

19 windowsize 19 ## Not run: data(clam1) shellx=clam1$x[38:70] shelly=clam1$y[38:70] window_shell=windowsize(shellx,shelly,1.8,9.4,0.2) ## End(Not run)

20 Index Topic datasets CLAM1, 3 climate, 4 elliot_yr1, 5 elliot_yr2, 6 otolith, 9 rwp_limpet, 16 Topic misc error.bar, 7 NextPow2, 8 plotproxy.error, 10 proxya, 12 proxyjk, 13 RESCALE, 15 SinMod, 17 windowsize, 18 Topic package ClamR-package, 2 rwp_limpet, 16 SinMod, 17 windowsize, 18 CLAM1, 3 ClamR (ClamR-package), 2 ClamR-package, 2 climate, 4 elliot_yr1, 5 elliot_yr2, 6 error.bar, 7 NextPow2, 8 otolith, 9 plotproxy.all (plotproxy.error), 10 plotproxy.all2 (plotproxy.error), 10 plotproxy.error, 10 plotproxy.error11 (plotproxy.error), 10 plotproxy1 (plotproxy.error), 10 proxya, 12 proxyjk, 13 RESCALE, 15 20